From Dr. Roy Spencer’s Global Warming Blog
2026 by Roy W. Spencer, Ph. D.
NOTE: I’ve made a couple of updates/clarifications below based upon comments on this post.
Lake Mead behind Hoover Dam is now within several feet of the lowest level at which turbine power generation must be greatly curtailed. Recent news articles often blame the cause on a “millennium drought” being experienced in the West. Of course, much of the western U.S. experiences drought conditions every summer, which is part of the normal climate there. Tens of millions of people like to live where it is sunny most of the year, and agriculture thrives where photosynthesis (lots of sunshine) is abundant… assuming enough water supply for irrigation.
Obviously, people need water, which is why virtually no one lives in the Sahara Desert. Add in the beauty of the Pacific Ocean, and it was inevitable that water demand would exceed supply. Even in 1922, before most people moved out west, the Colorado River Compact was negotiated to divide up the Colorado River’s water among seven southwestern states: California, Nevada, New Mexico, Arizona, Colorado, Utah, and Wyoming.
Most of the water in the Colorado River is from snowmelt in the Rocky Mountains. Most (50% to 70%) goes to agriculture. And, yes, the annual runoff flows into Lake Mead (which was filled after construction of Hoover Dam was mostly completed in 1935) have been below “normal” in recent years. But calling recent reduced streamflow (and as a result, lower Lake Mead water levels) the result of a “millennium drought” is sloppy reporting.
Numerous studies suggest that the original Colorado River Compact of 1922 was based upon a period of unusually high precipitation. This was discussed in a paper entitled “New Streamflow Reconstructions for the Upper Colorado River Basin: Placing Recent Droughts into a Centuries-Long Context“, which is a very readable review of various studies that have reconstructed Colorado River flows at Lee Ferry over the last 500 years from tree ring data. While the paper is a little dated (2006), the conclusions remain the same today.
Here are a couple of noteworthy quotes from that paper (bold added):
“The Stockton and Jacoby reconstruction of the annual flows from 1520 to 1961 contained several important findings, including that the highest sustained flows in the entire record occurred in the first part of the 20th century, the period upon which the Colorado River Compact of 1922 was negotiated, while the most persistent and severe drought occurred in the late 16th century.“
“…Colorado River allocations were based on one of the wettest periods in the past five centuries, while droughts more severe than any 20th-21st century event have occurred in the past.”
I’ve created a chart of the annual streamflows at Lee Ferry with some annotations showing periods of 3+ year droughts, the long-term trend (5th order polynomial fit) since stream gage data started in 1906, and four different estimates of the long-term (500-year) normal streamflow at Lee Ferry reconstructed from tree ring data.(UPDATE: the 2026 estimate is not just a repeat of the 2025 measurement, it is the central estimate of several quantitative evaluations of what the 2026 water year [October through September] will be based upon Lee Ferry gage measurements.)
As can be seen, the period before Hoover Dam was constructed was unusually wet. From the 1930s onward, there has been little net trend with some very wet years and very dry years, with up to a factor of 4-to-5 variations in the annual flows into Lake Mead. Six drought periods of at least 3 consecutive years are shown in red. Note that it took only 5 years to completely fill Lake Mead, which was accomplished by reducing the streamflow below the dam by about 30% to 40% of what the normal flows have been in recent decades.
From the above graph, I think one would be hard pressed to say there has been a statistically significant decrease in streamflow since the 1930s. (UPDATE: the linear trend for 1932 [a wet year] to 2025 is -0.013 +/- 0.016 million acre-feet per year, so it is not significant at the 68% confidence level). The major problem is that water usage exceeds what Mother Nature can reliably provide on a long-term basis. Several years ago I discussed this problem in a post in which I included the following graph (also rather dated) of Colorado River water supply versus usage.
As can be seen, since about 2000, usage has exceeded supply, and given the natural multi-year variations in streamflow, it was only a matter of time before Lake Mead water levels would fall to operationally low levels.
In that same post I showed the average of over 100 CMIP6 climate model runs that show precipitation in the Colorado River watershed increasing substantially with global warming, so blaming the recent dry years on human-caused climate change seems a stretch. (UPDATE: While it is true that drought is a function of both precipitation and evaporation, and warming can cause drought even if precipitation does not change, water releases from Hoover Dam have been declining since 1998, so increased evapotranspiration from irrigation uses downstream of the dam [the main use of Colorado River water] cannot explain decreasing Lake Mead water levels. Direct evaporation from the lake surface is very small and changes little with warming. And there has been no statistically significant trend in streamflow input at Lee Ferry since 1932.)
And calling the recent dry years a “millennium drought” is just fake news
